Vision for Future Innovative
Technology for Radiation
Oncology
D.A. Jaffray
Princess Margaret Cancer Centre
Techna Institute/Ontario Cancer Institute
University Health Network
University of Toronto
Toronto, Ontario, CANADA
ASTRO-AAPM-NIH’13
Disclosure
Presenter has a financial interest in some
of the technology reported here and
research collaborations with Elekta,
Philips, IMRIS, Varian, Quanta, and
Raysearch.
Results from studies using investigational
devices will be described in this
presentation.
Technology for Innovation in Radiation
Oncology: A Summary of the 2013 NCI
ASTRO AAPM Workshop
• Main Themes
– Innovative Technologies
– Advances in Imaging for Quantitative
and Validated Treatment Design
– Oncology Informatics
– Evidence Building
1
Capacity to Integrate
Molecular/Functional
Imaging in RT is
Significant
Database of
Dose Targets and
Tolerances
Calc’n
CT
Contouring
;
Optimization
CT/PET
MR
IMRT Beam
Patterns
Geometry Change
d
Specific Patient
Biology Change IGRT Adjustments
RT needs robust information
about the extent of disease.
Daisne et al Radiology 94 (2004)
Need to engage
pathology more in
target
characterization.
Too few investigators working on this problem.
Caution: Is someone hiding
delineation uncertainties in the PTV
margin?
PTV
2
From Anatomical to Functional
• Classical Dosimetric Objective:
– Deliver a uniform dose to the ‘clinical target’
– Minimize or limit the dose to surrounding normal
structures
The development of image-guidance
technologies addresses the geometric
challenges and enables pursuit of
functional or molecular sub-targets.
• Biological Objective:
But what is a
valid biological
target?
- Maximize therapeutic ratio
Conceptual Framework for Integration of
Functional/Molecular Imaging
“Incremental to the
concept of gross,
clinical, and planning
target volumes (GTV,
CTV, and PTV), we
propose the concept
of “biological target
volume” (BTV) and
hypothesize that BTV
can be derived from
biological images and
that their use may
incrementally improve
target delineation and
dose delivery.” - Ling
et al.
Ling et al., Int. J. Radiation Oncology Biol. Phys., Vol. 47, No. 3, pp. 551–560, 2000
From the ‘3D Hypothesis’ to the
‘BTV and 4D Hypotheses’
• BTV Hypothesis: Patterning radiation dose
according to imaged functional or molecular
distributions will increase the TR.
– A.k.a. ‘Biologically Targeted Radiation Therapy’
• 4D Hypothesis: Adapting to imaged changes
in geometry or function during RT will
improve the therapeutic ratio.
– A.k.a. ‘Adaptive Radiation Therapy’
3
Functional and Molecular Imaging for RT
• Tumour burden, altered metabolism, and clonogen
density (e.g. FDG, MRS)
• Tumour hypoxia (e.g. F-MISO, I/FAZA, CAIX, MRBOLD, HX4)
• Tumour proliferation (e.g. FLT)
• New imaging targets (e.g. FACBC amino acid, EGFR
for re-population)
• Functional imaging of crucial healthy tissues (e.g.
SPECT/CT/MR derived lung perfusion)
• Vascular and physiological measures (DCE-MR/CT,
MR DWI/ADC)
Adapted From ‘Theragnostic imaging for radiation oncology: dose-painting by
numbers’ - S.M. Bentzen - Lancet Oncol 2005; 6: 112–17
Impact of Specific and Sensitive Imaging of
Disease on Radiation Therapy
1. Reduce observer-dependent variation
in the extent of gross and clinical
targets.
2. Enable biologically-modulated targeting
of the radiation dose.
3. Enable prediction of response based
upon pre- or intra-treatment changes
in the image-based biomarkers.
See Steenbakers 2006, Bentzen 2005, Mayr 2010
Conceptually simple, but execution
has been very slow...
• There are signs of progress:
– MRI. Dominant lesion in multi-focal disease
(e.g. DCE/DWI/BOLD definition of dominant
lesion in prostate cancer – correlation with
pathology)
– FDG-PET. Targeting regions of elevated
metabolism as characterized by FDG-PET
– Hypoxia. Targeting regions of hypoxia as
characterized by F-MISO/FAZA/ATSM agents
• Challenges:
– Validation of image signal, cost of imaging/
pathology correlation, prescription/radiobiology,
technical (segmentation, robust imaging, motion
compensation, non-specific signal e.g. bladder)
4
Effect of Ignoring 4D in FDG-PET for the Lung
From the ‘3D Hypothesis’ to the
‘BTV and 4D Hypotheses’
• BTV Hypothesis: Patterning radiation dose
according to imaged functional or molecular
distributions will increase the TR.
– A.k.a. ‘Biologically Targeted Radiation Therapy’
• 4D Hypothesis: Adapting to imaged changes
in geometry or function during RT will
improve the therapeutic ratio.
– A.k.a. ‘Adaptive Radiation Therapy’
Computational Advances Needed for
Testing the ‘4D’ Hypothesis
Autosegmentation
Deformable
Registration
Dose
Tracking
Replanning
Time-based data architecture for adaptive and response studies.
5
Many scientific questions in the
domain of biologically-guided RT.
• What treatment targets are optimal for different tumor
histologies and genotypes?
• How to translate from radiobiological heterogeneity to dose
heterogeneity?
• What biological changes occur in tumor and normal tissue
during therapy and how do we adapt RT to those changes?
• How often do we need to adapt the plans?
• What other biomarkers beyond and in combination with
imaging do we need?
• How to integrate radiation therapy with other treatment
modalities to increase efficacy? How do we integrate all
these RT and other biomarkers?
Technology for Innovation in Radiation
Oncology: A Summary of the 2013 NCI
ASTRO AAPM Workshop
• Major undertones
– Need disruptive technologies in RO
– Need technical, quality, and safety
performance of these technologies
– Need process automation and oncology
informatics
A balance between disruptive technologies
and the need for evidence of efficacy and
effectiveness
Medical Physicists are capable of
inventing, developing, and deploying
technology in the clinical setting.
We benefit from a rich array of technology development
around us.
6
Next Gen RT Technologies:
Enabled by Better Dose Deposition, Robotics,
Computation, Imaging
Edmonton
Solution
Utrecht Solution
Protons+
PET-guided
RT
Viewray Solution
Adjacent Solutions
MR-Guided RT
Better dose control, faster imaging, higher CNR,
more responsive in delivery.
Technological Innovation in RT
• From Anatomic to Functional Targeting
– Validation for Target Delineation
– Anatomical and Response Adaptation
• Nanotechnology (hot and cold)
– Modification/Prediction of Response
– Dosimetry of drug delivery
• A New Breed of Multi-modal Machines
– Fast Physics that ‘Bury the Complexity’
• Our Biggest Treatment Machine Yet
– Standardization, Automation, Quality,
Efficiency, and Evidence
ASTRO-AAPM-NIH’13
7
An Apparent Conflict
Radiation Oncology and
Medical Physicists are
known for creation and
adoption of technology.
vs
Technology needs
to be better
evaluated.
Radiation Oncology and
Medical Physicists need
to be bridled to assure
we have evidence.
An Apparent Conflict
Opportunity
Radiation Oncology and
Medical Physicists are
known for creation and
adoption of technology.
vs
Technology needs
to be better
evaluated.
Radiation Oncology and
Medical Physicists need to
turns their creativity to
build evaluation into
innovation.
RT: A Highly Personalized Cancer Medicine
8
A Multi-modal Perspective
Uni-modal biomarkers are not sufficient for
informing clinical decisions and the full power
of genomics will be only be fully unleashed
through integration with multi-modal datasets.
Hypoxia, Receptor,
Permeability
Genetic/Proteomic/
Receptor
H&N
Image-based
Biomarkers
Tissue-derived
Biomarkers
OS<- HPV,p16: 20% (Shi 2009),
DFS<-FAZA-PET: 33% (Mortensen 2012)
LRC<-RT Compliance: 20% (Peters 2012) Intervention
Prostate
bPSA<IGRT: 25% (deCrevosier 2005),
bPSA<pO2: 25% (Milosevic 2012),
bPSA<Metformin: 10% (Zannella)
Performance
Outcome
Measures
RT, Sx, Cx
IGRT, IGS, IGDD
Clinician, Patient
Reported
NSCLC
.
LC<-FDG-PET Response: 40% (DeRuyescher, 2012)
OS<-CT Feature: 10% (Aerts 2013)
OS<-RT Technique: 25% (Liao, 2010)
RadPneum<-Cardiac Fn: 2.6 OR (Nalbantov, 2013;)
25
Personalized Cancer Medicine (PCM)
‘The right care for the right patient at
the right time.’
Patients or
practitioners?
"If it were not for the great
variability among individuals,
medicine might as well be a
science, not an art."
Sir William Osler, 1892
Our Biggest Machine Yet.
9
Paradoxically, Getting Personal Requires
Getting Industrial
Hypoxia, Receptor,
Permeability
Genetic, Proteomic,
Receptor
Image-based
Biomarkers
Tissue-derived
Biomarkers and
‘Omics
Intervention
Performance
RT, Sx, Cx
IGRT, IGS, IGDD
All three factors
characterized +
outcome
measures
Understanding cancer, developing personalized cancer medicine
strategies, and delivering high performance cancer therapy are
highly dependent activities.
ASTRO-AAPM-NIH’13
Medical Physics Efforts Toward
‘Industrial Medicine’
• Safety and Quality as Priorities
– Variance Tracking, Measurement
• Advancing a Nomenclature for RT
– ICRU 50, 62, 83
• Standardization in Treatment Methods
– Delineation Standards
• Measuring Outcomes
– Pay for Performance, Expertise Growth
• Engineering Principles
– FMEA, Control Charts, Measure and Correct
Strategy
10
Getting Industrial
• Standards and Nomenclature
– Communication and Automation
• Precision and Accuracy
– in our Measurements
– in our Treatment Delivery
• Collaboration
– Statistical Strength through Multi-institutional
Studies
• Informatics
– Managing the known unknowns and watching
for the unknown unknowns
Are Quality and Innovation Competitive?
ASQ (American Society for Quality) is the world’s leading authority on quality
and sole administrator of the Malcolm Baldridge National Quality Award.
ASQ 2010
Continuous Expertise/Knowledge
Development – Contribute and Benefit
Deasy et al. - Int J Radiat Oncol Biol Phys. 2010 March 1; 76
11
Converting on the Promise of
Personalized Cancer Medicine
• From delivering ‘state-of-the-art’ care to driving
the next generation of care.
– Radiation Oncology and Medical Physicists have
always innovated practice, but this needs to be
industrialized to accommodate the complexity of data
collection, decision making, and delivery.
• Maximizing intervention performance (quality) to
detect sub-populations and evaluate the value of
new, more personalized therapies
• Building cancer informatics tools to enable
analysis, exploration, and rapid evaluation of
novel therapies or stratification.
The Physicist’s Approach
• When you want to answer a really
big question…
–you get lots of talented people
engaged,
–you distribute the work,
–and you build a machine.
CERN
3000 Scientists
174 Universities
38 Countries
Working together on the border of France and Switzerland
Could translational cancer research become
another well-known large-scale science project?
12
Atlas Detector: Big Data Handling
•
•
•
•
•
•
>80 million Si pixels
>100,000 ionization ‘straws’
>100,000 calorimetry channels
25 ns resolution
>350,000 particles/sec/mm2
40 million bunch crossings/sec
Dedicated data system filters to
100,000/s (L1), 3000/s (L2), 200/s (L3
– for detailed analysis)
ASTRO-AAPM-NIH’13
What would the ‘machine’ for seeking
the limit of PCM look like?
• It’s design would push the limits of our science
– Drive new technologies and methods from basic science
• It would need to be precise
– Measurement/Terminology
• It should be accurate
– Standardized/Validated/Calibrated
• It should be efficient
– Affordable; large N; statistical power
• Needs to be integrated with healthcare delivery
Isn’t this this a remarkably good match for the skills in
Radiation Oncology and Medical Physics?
How would it work?
• A steady stream of well-characterized
patients colliding with a well-characterized
set of interventions.
• Surrounded by certified, calibrated beam
control and detection instruments
(controlled therapy, imaging, outcomes).
• Data streaming into an appropriately
staged real-time and off-line informatics
framework.
We just need to build it.
13
Summary
• Numerous emerging technologies with the
potential to impact Radiation Oncology
• Include advances in imaging,
nanotechnology, device development.
• Personalized cancer medicine is an
emerging medical, scientific, and technical
challenge and medical physicists in
radiation oncology have a significant role to
play.
• Advancing informatics as a central
research pillar for the field is timely and
appropriate.
14